Investigating fossil hydrothermal systems by means of fluid inclusions and stable isotopes in banded travertine: an example from Castelnuovo dell’Abate (southern Tuscany, Italy)

(1)

1 23

International Journal of Earth Sciences

GR Geologische Rundschau ISSN 1437-3254

Int J Earth Sci (Geol Rundsch) DOI 10.1007/s00531-015-1186-y

Investigating fossil hydrothermal systems

by means of fluid inclusions and stable

isotopes in banded travertine: an example

from Castelnuovo dell’Abate (southern

Tuscany, Italy)

Valentina Rimondi, Pilario Costagliola,

Giovanni Ruggieri, Marco Benvenuti,

Chiara Boschi, Andrea Brogi, Enrico

Capezzuoli, et al.

(2)

1 23

Your article is protected by copyright and

all rights are held exclusively by

Springer-Verlag Berlin Heidelberg. This e-offprint is

for personal use only and shall not be

self-archived in electronic repositories. If you wish

to self-archive your article, please use the

accepted manuscript version for posting on

your own website. You may further deposit

the accepted manuscript version in any

repository, provided it is only made publicly

available 12 months after official publication

or later and provided acknowledgement is

given to the original source of publication

and a link is inserted to the published article

on Springer's website. The link must be

accompanied by the following text: "The final

publication is available at link.springer.com”.

(3)

Int J Earth Sci (Geol Rundsch) DOI 10.1007/s00531-015-1186-y

ORIGINAL PAPER

Investigating fossil hydrothermal systems by means of fluid

inclusions and stable isotopes in banded travertine: an example

from Castelnuovo dell’Abate (southern Tuscany, Italy)

Valentina Rimondi1_{· Pilario Costagliola}2_{· Giovanni Ruggieri}1_{· Marco Benvenuti}2_·

Chiara Boschi3_{· Andrea Brogi}4_{· Enrico Capezzuoli}5_{· Guia Morelli}6_·

Massimo Gasparon6,7_{· Domenico Liotta}4

Received: 29 September 2014 / Accepted: 26 April 2015 © Springer-Verlag Berlin Heidelberg 2015

age of this circulation. The focus of the study was on fluid inclusions, rarely considered in travertine studies, but able to provide direct information on the physico-chemical char-acteristics of the original fluid. Uranium–thorium geochro-nological data provided further constraints on the: (1) age of tectonic activity; (2) age of the hydrothermal circulation; and (3) evolution of the Monte Amiata geothermal anom-aly. Results indicate that brittle deformation (NW- and SE-trending normal to oblique-slip faults) was active during at least the Middle Pleistocene and controlled a hydrothermal circulation mainly characterized by fluids of meteoric ori-gin, and as old as 300–350 ka. This is the oldest circula-tion documented to date in the Monte Amiata area. The fluid chemical composition is comparable to that of fluids currently exploited in the shallow reservoir of the Monte Amiata geothermal field, therefore suggesting that fluid composition has not changed substantially over time. These fluids, however, have cooled by about 70 °C in the last 300–350 ka, corresponding to a cooling rate of the Monte Amiata geothermal area of about 20 °C 100 ka−1_.

Keywords Banded travertine · C and O isotope geochemistry · Fluid inclusions study · Monte Amiata geothermal field

Introduction

The term “terrestrial carbonates” encompasses a wide spectrum of lithotypes mainly originated from calcium bicarbonate-rich waters under subaerial conditions, in a large variety of depositional and diagenetic settings (Flügel

2004). According to several classifications, the term “trav-ertine” indicates continental limestone deposited from non-marine, supersaturated calcium bicarbonate-rich waters, Abstract Southern Tuscany (Italy) hosts geothermal

anomalies with associated widespread CO2 gas-rich mani-festations and active travertine-deposing thermal springs. Geothermal anomalies have been active since the Late Miocene and have led to the formation of widespread Late Miocene–Pleistocene travertine deposits and meso- and epithermal mineralizations. This study investigates the travertine deposit exposed in the Castelnuovo dell’Abate area of southern Tuscany. Here, a fissure-ridge type trav-ertine deposit and its feeding conduits, currently filled with banded calcite veins (i.e. banded travertine), represent a spectacular example of fossil hydrothermal circulation in the peripheral area of the exploited Monte Amiata geother-mal field. The Castelnuovo dell’Abate travertine deposit and associated calcite veins were analysed to establish the characteristics of the parent hydrothermal fluids, and the

* Valentina Rimondi valentina.rimondi@unifi.it

1_{CNR - Istituto di Geoscienze e Georisorse, Via G. La Pira 4,}

50121 Florence, Italy

2_{Dipartimento di Scienze della Terra, Università di Firenze,}

Via G. La Pira 4, 50121 Florence, Italy

3_{CNR - Istituto di Geoscienze e Georisorse, Via Moruzzi 1,}

56124 Pisa, Italy

4_{Dipartimento di Scienze della Terra e Geoambientali,}

Università di Bari, Via Orabona 4, 70125 Bari, Italy

5_{Dipartimento di Fisica e Geologia, Università di Perugia,}

Via Pascoli, 06123 Perugia, Italy

6_{School of Earth Sciences, The University of Queensland,}

St Lucia, QLD 4072, Australia

7_{Australian National Centre for Groundwater Research}

and Training, The University of Queensland, St Lucia, QLD 4072, Australia

(4)

Int J Earth Sci (Geol Rundsch)

discharging to the surface from a deep geothermal system (Riding 1991; Ford and Pedley 1996; Fouke et al. 2000; Capezzuoli et al. 2014; Gandin and Capezzuoli 2014).

Southern Tuscany is characterized by a broad geother-mal anogeother-maly (Della Vedova et al. 2001) that is mainly cen-tred in the Larderello-Travale and Monte Amiata geother-mal areas (Batini et al. 2003) and strictly associated with CO₂ leakage and travertine deposits (Rogie et al. 2000; Minissale 2004). The geothermal systems of southern Tuscany mainly consist of two main reservoirs occurring at shallow crustal levels. The deeper reservoir (T of 300– 350 °C, depth 1000–3000 m) is hosted in metamorphic rocks, while the shallower reservoir (T lower than 200 °C, depth 500–1000 m) is hosted mainly in carbonate rocks (Duchi et al. 1992; Batini et al. 2003). Travertine deposits are the surface manifestation of hydrothermal fluid circula-tion within the carbonate reservoirs (Brogi et al. 2015 and reference therein) and are widely distributed in southern Tuscany and northern Latium (Minissale 2004, Brogi et al.

2010a, 2012; Capezzuoli et al. 2011). Over the last decade, ongoing research mainly focused on the inactive (i.e. fossil) depositional systems for neotectonic and palaeoseismologi-cal issues (Altunel and Karabacak 2005; Brogi and Capez-zuoli 2009; Brogi et al. 2010a, 2012, 2014; De Filippis et al. 2013; Uysal et al. 2007), as well as for palaeoclimate reconstructions (Manfra et al. 1974; Faccenna et al. 2008).

The study of fluid inclusions represents a unique tech-nique for the reconstruction of the physical and chemical characteristics of the travertine parent fluid and its deposi-tional environment. However, the application of this tech-nique to travertine deposits is rather new (Słowakiewicz

2003; Gibert et al. 2009; El Desouky et al. 2015) and has not been attempted before on Italian deposits. Fluid inclu-sions investigation in travertine is potentially complicated by: (1) the generally small size of the fluid inclusions (Pentecost 2005); (2) their metastability, which prevents the nucleation of the bubble upon cooling from the trap-ping conditions to room temperature (Shepherd et al. 1985; Diamond 2003); (3) the intrinsic difficulties in performing microthermometric analysis in calcite, which is often sub-jected to anelastic stretching phenomena (Roedder 1984; Shepherd et al. 1985), that may lead to misinterpretation of the results. This study is the first successful application of the fluid inclusions methodology to calcites, formed in the inactive (i.e. fossil) Pleistocene hydrothermal system of Castelnuovo dell’Abate in southern Tuscany (Fig. 1a, b). Here, a spectacular network of banded calcite veins cross-cutting both a fissure-ridge type travertine deposit and its substratum is exposed in an abandoned quarry. The traver-tine was further characterized in terms of its mineralogy, geochemistry, C and O isotopic composition, and age using U-series dating. The results provide information on the age of the tectonic activity and hydrothermal circulation.

Combined with previous studies on active and fossil neigh-bouring hydrothermal systems (Frondini et al. 2009; Gas-parrini et al. 2013), these results provide new insights into the evolution of the shallow hydrothermal reservoir of the Monte Amiata region and its possible cooling rate during the last 300–400 ka.

Geological setting

The Monte Amiata Middle–Late Pleistocene volcanic complex (Ferrari et al. 1996; Cadoux and Pinti 2009) is located in the inner Northern Apennines (Fig. 1a), a Ter-tiary (Late Cretaceous–Early Miocene) belt derived from the convergence and collision between the Adria micro-plate and the European micro-plate (Molli 2008 and references therein) and its interplay with the opening of the Tyr-rhenian Basin (Bartole 1995). After the stacking of the tectonic units, extensional tectonics (Carmignani et al.

1994; Brogi et al. 2005) and coeval magmatism (Pec-cerillo 2003; Dini et al. 2005) affected the previously overthickened continental crust since the Early–Middle Miocene (Brunet et al. 2000) and led to the exhumation of deeper tectonic units (Carmignani et al. 1995), wide-spread geothermal anomalies (Della Vedova et al. 2001) and hydrothermal circulation (Tanelli 1983). Around Monte Amiata, hydrothermal circulation was responsible for the world-class Hg and Sb deposits, extensively mined up to thirty years ago (Rimondi et al. 2015), as well as to barren hydrothermal calcite veins (Gasparrini et al. 2013). Hydrothermal fluid flow is still active in this area as mani-fested by the travertine deposition from thermal springs at Bagno Vignoni and Bagni San Filippo (Fig. 1a), the occurrence of numerous CO2-rich gas emissions (Fron-dini et al. 2009) and the two geothermal fields (Bagnore and Piancastagnaio) exploited for electricity production (Batini et al. 2003). All these features can be related to the thermal anomaly generated by a granitoid cooling at about 6–7 km below sea level (Gianelli et al. 1988; Ber-tini et al. 1995; Brogi 2008).

Hydrothermal circulation in the Monte Amiata area (i.e. Hg mineralization, thermal springs, travertine deposits and gas emissions) has mainly been controlled by normal faults and their associated transfer zones, trending NW and NE, respectively (Brogi et al. 2010b). Similarly, at Castelnuovo dell’Abate, the travertine deposit is aligned with a N50° trending fault (Fig. 1b), which is interpreted as a transfer zone and played the role of the main conduit that chan-nelled the feeding hydrothermal fluids. This fault separates Late Oligocene–Early Miocene quartz-feldspar sandstone (Macigno Fm) belonging to the Tuscan Nappe, from the overlying Cretaceous marl and marly limestone (S. Fiora Fm) belonging to the Ligurian units (Fig. 2).

(5)

Castelnuovo dell’Abate travertine and banded calcite vein system

The Castelnuovo dell’Abate travertine depositional system crops out about 300 m south of the homonymous village (Fig. 1b) and can be described as a fissure-ridge deposi-tional system (Fig. 3). Travertine and related banded calcite were intensively exploited for ornamental use for centuries, and this made it possible to study their general features on the faces of the many abandoned quarries. The whole trav-ertine body was affected by brittle deformation that caused intense fracturing and faulting.

According to Altunel and Hancock’s (1993) classifica-tion, two main depositional associations have been rec-ognized: bedded and banded travertine (Fig. 3), the latter consisting of: (1) banded calcite veins filling the vertical, straight-to-sinuous fissure crossing the central part of the (eroded) fissure-ridge body (Fig. 3), and (2) sub-horizon-tal to sub-vertical banded calcite veins intruded within the substratum (limestone and marl of the S. Fiora Fm and conglomerate), often showing mutual crosscutting

relationships due to hydrofracturing (Fig. 4). Banded cal-cite veins represent most of the travertine body and con-sist of crystalline, laminated crusts (Fig. 5a) precipitated in hypogean conditions. These are formed by onyx-like, well-banded, centimetre-to-metre-thick, white-to-brown calcite bands (Fig. 5b). Bands are formed by palisade, fibrous or blocky prismatic crystals grown in syntaxial continuity (Fig. 5a). Sub-horizontal veins follow closely the geometry of the beds of the hosting limestone and marl forming the travertine substratum, like sills (Fig. 4d–f). Pinched (char-acterized by a V-shaped morphology; Fig. 5c) and highly brecciated bands are locally present and represent the evo-lution of superimposed fractures that changed their trajec-tories isolating lithons of previously deposited bedded trav-ertine. Occasionally, centimetre-scale breccia fragments, cemented by interstitial calcite including light grey carbon-atic fragments derived from the host, have been incorpo-rated within the banded veins. The thickness of the banded calcite veins is extremely variable, ranging from some cen-timetres to a few decimetres in the travertine substratum. In contrast, their thickness varies from some millimetres to a Fig. 1 a Geologic map of

southern Tuscany and main geothermal features; the present-day travertine deposits associated with active ther-mal springs are also indicated (Bagno Vignoni and Bagni San Filippo); b geological map of Castelnuovo dell’Abate and sur-rounding area

(6)

Fig. 2 Tectono-stratigraphic units in the study area (modified after

Batini et al. 2003; Gasparrini et al. 2013); M–P–Q: Miocene, Plio-cene and Quaternary sediments; MR—magmatic rocks; Tuscan Nappe (TN): TN3—Early Miocene–Cretaceous clayey-turbiditic suc-cession; TN2—Cretaceous–Rhaetian carbonate succession, TN1— Late Triassic evaporite succession; Monticiano–Roccastrada Unit (MRU): MRU3—Triassic Verrucano Group; MRU2—Palaeozoic Phyllite–Quartzite Group; MRU1—Palaeozoic Micaschist Group; GC—Palaeozoic Gneiss Complex. Tuscan Nappe: Ev—Late Triassic evaporites (Burano Fm) composed of an alternation of dolostone and dolomitic limestone and gypsum layers often brecciated; Cr—Late Triassic carbonate succession (Calcari a Rhaetavicula contorta Fm) composed of Rhaetian dark limestone interbedded with decimetre thick marl and marly limestone; Cm—Early Jurassic massive grey limestone (Calcare Massiccio Fm); Cra—Early Jurassic red

nodu-lar limestone (Calcare Rosso Ammonitico Fm); Cs—Early Jurassic marly limestone and grey cherty limestone (Calcare Selcifero Fm); Mp—Middle Jurassic marl and marly limestone (Marne a Posidonia Fm); Di—Late Jurassic radiolarite (Diaspri Fm); Sc—Cretaceous– Oligocene shale, marl, limestone, calcarenite and calcirudite (Scaglia Toscana Fm); Ma—Late Oligocene–Early Miocene sandstone and shale (Macigno Fm). Ligurian Units: CU—Eocene–Oligocene lime-stone and shale (Canetolo Fm); Sf—Cretaceous limelime-stone, marl and shale (S. Fiora Fm); Pf—Cretaceous sandstone (Pietraforte Fm); Mm—Palaeocene–Eocene marl and limestone (Monte Morello Fm); Pe—Jurassic peridotite; Ga—Jurassic gabbro; Ba—Jurassic basalt; Ra—Late Jurassic radiolarites (Diaspri Fm); Cc—Early Cretaceous siliceous limestone (Calcari a Calpionella Fm); Cp—Early Creta-ceous shale and siliCreta-ceous limestone (Calcari a Palombini)

Fig. 3 a Panoramic view of some quarry faces showing part of the architectural components of the travertine fissure-ridge deposit; b detail of

(a); banded travertine vein and opposite dipping walls made by bedded travertine can be recognized

(7)

few metres (maximum values of about 3 m) in the central part of the fissure ridge, along the main trace of the fault.

The bedded travertine consists of scattered out-crops, in some cases representing metre-sized lithons embedded within banded calcite veins. It comprises variously inclined, wavy-crinkled laminated to non-stratified lithofacies deposited in epigean, proximal-to-distal conditions by the thermal waters flowing from the vent(s) along slopes (Fig. 5d). According to Gandin and Capezzuoli (2014), different types of crusts and granular facies with variable thickness and lateral evo-lution can be recognized (Fig. 5e, f). Such lateral facies organization, typical of a thermal depositional system (Guo and Riding 1998, 1999), provides evidence of dif-ferent microenvironments (pools and shallow basins,

microterraced slopes) repeatedly following one upon another.

Methods

Banded travertine was collected both in the substratum of the travertine body and in the travertine ridge (Figs. 4d, 5c) with the aim to characterize fluid circulation inside the trav-ertine system. The same set of banded travtrav-ertine samples (TRS1, TRS2/1 and TRS2/2) was analysed for fluid inclu-sions, mineralogy, C and O isotopes, and U-series dating. Materials for isotope analyses and dating were obtained from single bands by microdrilling. To this purpose, only bands that displayed a width larger than the drill tip were Fig. 4 a Banded calcite vein crosscutting the substratum of the

fis-sure-ridge travertine deposit, made by conglomerate; b sub-vertical centimetre-to-decimetre-thick banded calcite veins crosscutting the substratum of the fissure-ridge travertine deposit, made by marl and clayey marl belonging to the Cretaceous S. Fiora Fm; c network of millimetre-thick banded calcite veins crosscutting a previous calcite

vein, related to hydrofracturing; d sub-horizontal banded calcite veins (sill-like) parallel to limestone beds belonging to the Cretaceous S. Fiora Fm and location of sample TRS1; e sub-horizontal banded cal-cite veins (sill-like) with syntaxial geometry, parallel to marl beds belonging to the Cretaceous S. Fiora Fm; f crosscutting relationships between different generations of banded calcite veins

(8)

selected. Different bands were analysed for stable isotopes and U-series dating. Although these bands appear homo-geneous at the scale of the hand specimen, it cannot be

excluded that each sample represents a mixture of differ-ent depositional micro-evdiffer-ents during travertine deposition. A more detailed investigation at the microscale was beyond Fig. 5 a Typical aspect of banded travertine, formed by laminated,

differently coloured calcite and with syntaxial geometry veins; b detailed view of a banded calcite veins, formed by mm-to-cm, pris-matic-to palisade crystal calcite bands and characterized by different colors; c common feature of the banded travertine mutually crossing

with V-shape aspect; location of samples TRS2/1 and TRS2/2 are reported; d alternation of crystalline crusts and detrital, sandy depos-its documenting a slope environment; e encrusted reeds in horizontal position from a distal slope environment; f bedded travertine: super-position of paper-thin raft of pool environment

(9)

the scope of the present work, since the main goal was to obtain general information on the fluid characteristics and on the age of the hydrothermal system.

Fluid inclusions were studied in 70- to 100-µm-thick double-polished wafers of calcite samples. Particular care was taken during sample preparation to avoid or minimize anelastic calcite stretching; a low-speed trim saw was used for sample cutting, and little pressure was applied on the samples during grinding and polishing (Goldstein and Reynolds 1994).

Microthermometric measurements were performed using a Linkam THMSG600 heating–freezing stage cou-pled with a microscope Ortholux II POL-BK (Leitz) at the CNR-IGG in Firenze. The stage was calibrated by using pure H₂O with critical density and mixed H₂O–CO₂ (CO₂ 25 % M) synthetic fluid inclusions. Accuracy was estimated to be ±0.2 °C for final ice melting temperature (Tmice) measurements and ±0.1 °C for homogenization tempera-ture (Th). Due to the inelastic behaviour of calcite, heating generally preceded the freezing stage to avoid decrepitation and stretching phenomena commonly associated with vol-ume expansion during ice formation (Roedder 1984). The temperature of homogenization was systematically meas-ured only on liquid-rich inclusions. In vapour-rich inclu-sions, no remarkable increase in the liquid/vapour ratio was observed within the Th range of liquid-rich inclusions; thus, heating was stopped at 160 °C to overcome overheat-ing phenomena and consequent decrepitation of liquid-rich inclusions.

The temperature of final ice melting was determined in liquid-rich inclusions, while Tm_ice was generally not observed in vapour-rich inclusions. The apparent salin-ity of fluid inclusions (Hedenquist and Henley 1985) was calculated from Tmice using the equation of Bodnar and Vytik (1994) and expressed in wt% NaCl eq. To assess the presence of non-condensable gases in the inclusions, a number of crushing tests was carried out (cf. Roedder

1970).

SEM–EDS analyses were carried out at the Centro Inter-dipartimentale di Microscopia Elettronica (MEMA, Fire-nze) on polished sections using a SEM/EDS ZEISS MA 15 and an acceleration potential of 25 kV.

Single bands were microdrilled along specific transects of calcite veins, orthogonal to vein direction, powdered if necessary and analysed for C and O isotopes. Calcite sam-ples were analysed for δ13_{C and δ}18_{O by using a} Finnigan-MAT 252 (CNR-IGG in Pisa) after dissolution in phos-phoric acid by using a common-acid bath technique at 70 °C (based on McCrea 1950). Data were corrected for the usual isobaric interferences following the procedure of Craig (1957) modified for a triple-collector instrument. The δ18_{O value was calculated considering the acid} frac-tionation factor and calibrated relative to the laboratory

calcite standard (Friedman and O’Neil 1977). Carbon and oxygen isotope abundances are reported in δ values in per mill relative to the Vienna Peedee belemnite (V-PDB) and Standard Mean Oceanic Water (SMOW), respectively. The internal standards were measured at regular intervals within the sample set, and the uncertainty was less than 0.20 and 0.10 ‰ for δ13_{C and δ}18_{O, respectively.}

For U-series dating, samples were washed with deion-ized water, cleaned in an ultrasonic bath to remove any pos-sible contamination from the surface and dried at 40 °C on a hotplate. A total of five calcite bands were microdrilled using a bench drill, two bands from sample TRS1 (at 12 and 20 cm distance from the wall, labelled TRS1_12 and TRS1_20, respectively, the notation is applied to the other samples), one vein from sample TRS2/1 (at 25 cm from the wall) and one vein from sample TRS2/2 (at 35 and 90 cm from the wall). U–series dating was carried out in the Radiogenic Isotope Laboratory at the School of Earth Sci-ences (The University of Queensland) using a Nu Plasma multi-collector inductively coupled mass spectrometer (MC-ICP-MS). Samples (about 0.01 g powders) were com-pletely dissolved using ultra-pure double-distilled concen-trated HNO₃ mixed with a 229_Th–233_{U spike. Hydrogen} per-oxide was added to the samples after digestion to remove any possible organic matter, and a new anion-exchange column chemistry procedure (modified after Edwards et al. 1987; Zhao et al. 2001; Clark et al. 2012) was used to purify U and Th. Unlike the procedure commonly used in the past, in this modified procedure, U and Th are eluted together using ~3 ml of a 1 % HNO3 + 0.03 % HF mixture. The mixed solution was then directly injected through a DSN-100 desolvation system for simultaneous U and Th isotope measurement following the procedures first described in Zhou et al. (2011). Correction of initial 230_{Th was done using a two-component correction scheme.} The non-radiogenic 230_{Th was corrected assuming a} bulk-Earth atomic 230_Th/232_{Th ratio of 4.4 ± 2.2 × 10}−6_{, while}

230_Th/238_{U and}234_U/238_{U activity ratios of the samples were} calculated using the decay constants given in Cheng et al. (2000). Final U–Th ages were calculated using the Isoplot/ Ex 3.0 Program (Ludwig 1999).

Results

Mineralogy

X-ray diffraction analysis (XRD) shows that the banded travertines are mainly composed of calcite (~90 wt%), and to a minor extent by phyllosilicates and quartz (~10 wt%). No aragonite was detected, suggesting that this mineral represents less than 5 %, if present. In transmitted light microscopy, calcite is commonly present as elongated

(10)

crystals fractured perpendicular to their accretional direc-tion (Fig. 6a), or more rarely as small (around 50–100 µm) crystals with a blocky texture (Fig. 6b). Abundant opaque minerals mark the accretional margins of calcite, resulting in a typical “V” morphology in thin section (Fig. 7a). These phases were identified by SEM–EDS analysis as pyrite, Fe-(hydr)oxides, and/or pseudomorph Fe-(hydr)oxides after pyrite (Fig. 7b; Table 1). Two calcite vein samples (TRS2/1 and TRS2/2) contain fluorite (Fig. 7c) in discrete accretional bands along calcite blocky crystals. Celestine (SrSO₄) is an additional accessory mineral (Fig. 7d) that occupies intercrystalline spaces between calcite crystals. Calcite is generally pure, but with some elongated bands rich in Sr (Sr ~ 3 wt%, based on SEM–EDS semi-quantita-tive analysis) (Fig. 7d), and associated with celestine. Arse-nic content is as high as 2–3 wt% in Fe-(hydr)oxides and pyrite (Fig. 7e; Table 1).

Fluid inclusions

Fluid inclusion types and petrography

In general, fluid inclusions are scarce in the examined calcite, although they were found in both elongated and blocky calcite crystals (Fig. 8a–d). At room temperature, fluid inclusions are two-phase (aqueous liquid plus vapour) liquid-rich (L1), vapour-rich (V) or more rarely only liquid (L2) (Fig. 9a–f). The absence of a vapour phase in the L2 inclusions probably results from failure of bubble nuclea-tion due to metastability processes. Vapour-rich inclu-sions have variable vapour-to-liquid ratio and may occur as two-phase inclusions (vapour plus minor liquid; V1) or as apparently one-phase inclusions (vapour; V2), though a small amount of liquid, not visible under the microscope, may also be present. Groups of L1/L2 inclusions often show a three-dimensional random distribution, suggesting that they are primary in origin (Roedder 1984) and that can be considered fluid inclusion assemblages (FIA) accord-ing to Goldstein and Reynolds (1994). Single large size

(relative to host crystal) L1 inclusions can be also consid-ered primary in origin (Roedder 1984).

In elongated crystals, some inclusions are located in the accretion margins of calcite, describing the typical V-shape already mentioned for pyrite and Fe-(hydr)oxides (Figs. 7a, e, 9e), and indicating a primary origin according to the cri-teria of Roedder (1984). The size of the fluid inclusions ranges from 10 µm to 200 µm in the direction of the maxi-mum length (Figs. 7a, 8c, d).

In the accretion margin of calcite, L1 inclusions are often scarce (maximum two inclusions per single crystal), whereas large dark inclusions are more frequent (Fig. 9d). Most of the latter should be considered as opened inclu-sions that lost their fluid upon wafer preparation. Some sin-gle inclusions, however, also have a small portion of liquid (Fig. 9e), indicating that they are actually V1 inclusions. Rare groups of inclusions, consisting of a few L1 inclu-sions sometimes associated with L2, and single isolated L1 inclusions occur in the internal part of the elongated calcite crystals, as well within blocky calcite crystals (Fig. 9c–f).

L1 and L2 inclusions usually show irregular shape, although in some cases they are more rounded, or mimic the rhombohedric shape of the host crystal (Fig. 9c). The size of fluid inclusions ranges from 3 µm up to 200 µm in the direction of the maximum length. In some cases, neck-ing-down was suspected to have affected L1 inclusions. This process can explain the presence of the rare L2 inclu-sions coexisting with L1 incluinclu-sions.

Groups of L1/L2 inclusions are sometimes associated with large V1 and/or V2 inclusions in the internal part of the elongated calcite crystals (Fig. 9a). The genetic coex-istence of L1, L2, V1 and V2 in elongated calcite is likely related to fluid immiscibility processes resulting from boil-ing or effervescence. The variable vapour-to-liquid ratio showed by V1 inclusions is a typical consequence of the contemporaneous trapping of liquid and vapour within sin-gle inclusion (i.e. heterogeneous trapping) under immis-cibility conditions (Roedder 1984). The large single V1 and V2 inclusions in the accretion margins of calcite also Fig. 6 Photomicrographs (transmitted light, parallel nicols) of elongated (a) and blocky (b) calcite in the Castelnuovo dell’Abate travertine

(11)

Fig. 7 Mineralogy of the Castelnuovo dell’Abate travertines (sample

TRS2/2); a location of opaque minerals and fluid inclusions in the travertine thin sections (transmitted light microscopy, parallel nicols);

b pyrite (py), Fe-oxides (ox Fe), and quartz (qz) (SEM, backscattered

electron); c fluorite (transmitted light microscopy, parallel nicols);

d celestine and Sr-enriched elongated bands in travertine calcite

(SEM, backscattered electron); e detail of V-shaped opaque mineral assemblages of pyrite and Fe-oxides (SEM, backscattered electron). Point EDS analysis of the numbered phases are reported in Table 1

Table 1 SEM-EDS point

analysis of As-rich mineral phases (Fe-(hydr)oxides and pyrites) numbered in Fig. 7e

Si, Al, O and Ca in the mineral detection are attributed to the neighbouring areas Point analysis

(Fig. 2e)

Mineral phase Elements in wt%

O Si Al S Ca Fe As Total 1 Fe-(hydr)oxides 30.2 2.6 3.9 – 4.6 56.8 1.9 100.0 2 Fe-(hydr)oxides 29.6 1.1 – 3.4 2.1 60.9 3.0 100.0 3 Pyrite 6.7 – – 46.6 6.2 38.4 2.2 100.0 4 Pyrite – – 54.1 1.7 – 42.4 1.8 100.0 5 Fe-(hydr)oxides 30.2 2.6 3.9 – 4.6 56.8 1.9 100.0

(12)

testify the presence of a vapour phase resulting from fluid immiscibility.

Microthermometric data

The results of microthermometric analyses are summarized in Table 2. To avoid data collection from fluid inclusions affected by necking-down or heterogeneous trapping pro-cesses, we discarded the Th of FIA that showed rather vari-able Th values. Th and Tm_ice of single inclusions always fall within the Th and Tm_ice ranges of the FIAs; these data Fig. 8 Photomicrographs (transmitted light, parallel nicols) of fluid inclusions hosted in a, b blocky calcite, and c, d elongated calcite of the

Castelnuovo dell’Abate travertine

Fig. 9 Photomicrographs (transmitted light, parallel nicols) of

fluid inclusions hosted in banded calcite veins of the Castelnuovo dell’Abate travertine; a FIA made of liquid-rich (L1), vapour-rich inclusion (V1), and only vapour (V2) in elongated crystals; b, c

examples of L1 inclusions in elongated calcite crystals; d coexistence of a liquid-rich (L1) and only vapour (V2) inclusions in elongated calcite; e vapour-rich inclusion (V1) with a small portion of liquid; f typical FIA of blocky calcites made of two L1 inclusions

Table 2 Summary of the microthermometric data collected for

Castelnuovo dell’Abate travertine (L1 type of inclusions)

Numbers in brackets refer to the number of fluid inclusions analysed for each sample and calcite type

Sample Calcite type Th range (°C) Tm_ice range (°C) TRS1 Elongated 92.0/136.6 (14) −0.2/−0.6 (11) TRS2/1 Elongated 123.8/131.2 (5) −0.1/−0.7 (6)

Blocky 90.4/139.5 (35) _{0.0/−0.8 (8)} TRS2/2 Elongated 91.6/136.8 (15) _{0.0/−0.7 (14)}

Blocky 92.7118.8 (3) _{−0.5 (1)}

(13)

are reported together with those of the FIAs in the fre-quency histograms of Figs. 10 and 11. Most FIAs of L1 inclusions showed consistent Th data accordingly to the criteria of Goldstein and Reynolds (1994). Distinct FIAs are usually characterized by different Th ranges. The whole Th range of fluid inclusions, occurring both in elongated and in blocky calcite crystals, is comprised between 90 and 140 °C, with most of values comprised in the 105–140 °C range (Fig. 5).

Th of single L1 inclusions in blocky calcite as well as Th of L1 inclusions in elongated crystals fall within the Th range of L1 inclusions of the FIAs in blocky calcite,

suggesting that they trapped similar fluids. Some L2 inclu-sions nucleate a bubble after freezing; however, the Th measured after bubble nucleation was always higher than the Th of the coexisting L1 inclusions, likely because of stretching process upon ice formation at low tempera-ture. These data were not considered in the following discussion.

Decrepitation phenomena during cooling also reduced the number of L1 fluid inclusions analysed for Tm_ice (n = 40), which varied from −0.8 to 0.0 °C in both cal-cite types (Fig. 11), corresponding to salinities of 0–1.4 wt% NaCl eq. Apparent salinities are homogenously dis-tributed among the analysed samples and do not show any clear correlation with Th (Fig. 12). During low-temperature microthermometric analysis, no ice melting was observed within V1 and V2 inclusions.

Fig. 10 Results of microthermometry: Th for fluid inclusions hosted

in elongated and blocky calcite

Fig. 11 Results of microthermometry: Tm_ice for fluid inclusions hosted in elongated and blocky calcite

Fig. 12 Th versus salinity for fluid inclusions hosted in elongated

and blocky calcite

Table 3 Isotopic composition of the single bands of calcite veins of

the Castelnuovo dell’Abate travertine deposit

TRS2/1a-c and TRS2/2a-e refer to bands microdrilled along the tran-sect of a calcite vein

Sample δ13C (PDB) δ18O (PDB) δ18O (SMOW) TRS1 0.9 _−11.6 19.0 TRS2/1 a 1.1 _−11.7 18.8 TRS2/1 b 1.5 −12.4 18.1 TRS2/1 c 0.8 −12.0 18.5 TRS2/2 a 1.4 −10.2 20.4 TRS2/2 b 1.4 −10.7 19.9 TRS2/2 c 1.8 −9.7 20.8 TRS2/2 d 2.2 −11.4 19.1 TRS2/2 e 1.0 _−11.8 18.7 TRS3 2.6 _−9.5 21.0

(14)

Crushing tests

Crushing tests were performed successfully only on small number of L1 inclusions. Upon crushing, few inclusions showed the contraction of the gas bubble, while in the other inclusions the gas bubbles expanded but did not completely fill the cavity of the inclusions. Bubble expansion indicates the presence of pressurized gases (>1 bar) in the bubble. These gases are probably made up mainly of CO₂, as this is the dominant gas in systems depositing travertine. An estimate of CO₂ content was made by using the graphical method of Sasada (1985) based on bubble behaviour dur-ing crushdur-ing tests. The CO₂ concentration is between 0.07 and 0.14 mol% considering a Th range of 100–140 °C for inclusions showing bubble expansion with incomplete cav-ity filling. For inclusions exhibiting bubble shrinkage, the CO₂ content is below 0.07 mol%.

Stable isotope composition

Oxygen and carbon isotope data of calcite veins are reported in Table 3. δ13_{C showed a consistently positive} value for all samples (average value = 1.5 ‰; 2σ = 0.6 %). Along transects, calcite veins showed very small fluctua-tions in oxygen and carbon composition (Table 3), sug-gesting minor (isotopic) variations in fluid geochemistry during vein deposition. δ13_{C and δ}18_{O of the Castelnuovo} dell’Abate calcite veins are similar to those observed

for travertines of central Italy (Fig. 13). At Castelnuovo dell’Abate, δ18_{O of calcite is relatively low but} compara-ble to those of the Bagni San Filippo travertine (Minissale

2004).

In contrast to other Italian deposits, the Castelnuovo dell’Abate travertine does not show a clear positive δ13 C-δ18O correlation (R2 = 0.37) (Fig. 14).

U/Th dating

U concentrations in the samples varied between 1.00 and 2.84 mg kg−1_{, while concentrations of}232_{Th ranged} between 18.07 and 1041 μg kg−1_{(average of 494.4 ± 1.15}

(2σ) μg kg−1_{) (Table}₄_{). The measured ratio for}230_Th/232_Th varied between 114.55 ± 0.51 and 16.72 ± 0.09 (± 2σ). After non-radiogenic or detrital 230_{Th correction, the} analysed vein samples provided an age range between 368 ± 112 ka and 389 ± 19 ka. Corrected ages of two bands from sample TRS1 were 385 ± 75 ka (TRS1_12) and 368 ± 112 ka (TRS1_20); 389 ± 19 ka for TRS2/1_25, and 370 ± 14 ka for TRS2/2_35 (Table 4). These ages are identical considering the uncertainties due to detrital 230_Th correction and define the samples maximum ages. In con-trast, sample TRS2/2_90 has a high U content and does not fit the U–Th evolution trend. This implies partial U loss, resulting in excess 230_{Th unsupported by}234_{U decay.}

Although the calibration method was designed to mini-mize the error caused by the uncertainty in decay constant for 230_{Th and}234_{U, the age of the samples is close to} secu-lar equilibrium and therefore strongly affected by error Fig. 13 δ13_C-δ18_{O composition of Latial and Tuscan travertines}

com-pared to the Castelnuovo dell’Abate deposit. Data from Fouke et al. (2000), Fritz (1965), Manfra et al. (1976) and Minissale (2004). Latial travertine—1 Tivoli, 2 Canino, 3 Viterbo, 4 Cisterna di Latina,

5 Fiano Romano, 6 Ferentino. Tuscan travertine—7 Sarteano,

8 Bagni San Filippo, 9 Rapolano, 10 Casciana Terme

Fig. 14 δ13_C–δ18_{O correlation in Castelnuovo dell’Abate travertines}

(this study) and other Tuscan and Latial deposits (data from Minissale et al. 2002; Fouke et al. 2000; Gonfiantini et al. 1968). Open square Castelnuovo dell’Abate, open pentagon Bagni San Filippo, grey filled

circle Castelnuovo Berardenga, filled star Bagnaccio, black filled

cir-cle Casciana Terme, filled triangle Massa Marittima

(15)

magnification. Another factor complicating the interpre-tation of the data is the uncertainty in the correction for a possible contribution of non-radiogenic (initial/detrital) 232_{Th, as the assumed bulk-Earth values may not be} appli-cable to these samples. Irrespective of these limitations, the data seem to suggest that (with the exception of sample TRS2/2_90, which cannot be accurately dated using this method) all the samples formed virtually at the same time at ~370–390 ka.

Discussion

Mineralogy of Castelnuovo dell’Abate travertine deposit

Banded travertine veins are generally composed by almost pure CaCO₃, which may occur in the form of calcite or aragonite, depending on numerous factors such as tem-perature, chemical composition (Mg/Ca ratio; Kele et al.

2011), presence of Sr2+_{or SO}

42− (Malesani and Vannucci

1975), pCO₂ and CO₂ degassing rate (Kele et al. 2008). It is generally assumed that aragonite forms preferentially at

T > 40 °C, while calcite dominates in the T < 30 °C domain for unstirred solution belonging to the H2O–CO2–CaCO3 system (Fouke et al. 2000). Actually, it has been noted, however, that changes in the degree of supersaturation due to CO₂ degassing greatly influence these temperature limits (Chafetz et al. 1991; Renaut and Jones 1997). The rate of precipitation may further control the polymorph being precipitated, with rapid deposition favouring arago-nite but not necessarily excluding calcite from the mineral assemblage (Renaut and Jones 1997). More precisely, slow degassing rates under low PCO2 favour calcite precipitation,

whereas aragonite is more frequent under high degassing rates and high levels of supersaturation (Arnórsson 1989). The Castelnuovo dell’Abate banded veins are essentially composed of calcite. Aragonite is absent, and there is no evidence of a post-depositional transition from aragonite to calcite. The typical radial texture of calcite observed in

the field and the supposed conditions of precipitation (i.e. during fluid immiscibility) are indicative of a high rate of precipitation of the carbonatic phase (Simmons and Chris-tenson 1994). Relatively high levels of Sr in calcite further confirm this hypothesis, since the co-precipitation of ele-ments with distribution coefficient <1, like Sr, is enhanced by high rates of precipitation (Kele et al. 2008).

Low PCO2 during fluid immiscibility might then have

been the controlling factor for the deposition of calcite along the Castelnuovo dell’Abate fault, consistent with the relatively low concentrations of dissolved CO₂ observed in fluid inclusions study (see “Mineralogy of Castelnuovo dell’Abate travertine deposit” section). Similar conditions have been documented in Kenya (Renaut and Jones 1997; Renaut et al. 2013) and in New Zealand (Jones et al. 1996), where calcite precipitates directly from waters at tempera-ture higher than 90 °C.

The Castelnuovo dell’Abate banded calcite veins dis-play distinctive mineralogical features, like the presence of fluorite, pyrite and associated high levels of As (Table 1). Travertine is often enriched in volatile and semi-volatile elements, such as F (Pentecost 2005). Fluorite has been documented in Colorado, New Mexico and Kenya traver-tines (Pentecost 2005; Renaut et al. 2013). Since fluorite shows prograde solubility in the temperature/salinity range suggested by fluid inclusions (Richardson and Holland

1979), fluorite likely followed calcite deposition at Castel-nuovo dell’Abate, being related to cooling of the hydro-thermal fluid (Renaut et al. 2013) and/or a drop in salinity, following fluid mixing. High levels of F (several mg L−1₎

are also documented in modern thermal springs of Tuscany and Latium (Minissale et al. 2002; Frondini et al. 2008), and travertine-depositing waters are often oversaturated in fluorite (Di Benedetto et al. 2011). The elevated concentra-tions of F and Sr in fossil and current travertine deposits suggest deep circulation of waters inside the Tuscan rock pile (Minissale et al. 2002). According to Barbieri et al. (1976), these fluids interact with the Late Triassic evaporite horizon (Burano Fm.) and the Mesozoic carbonate succes-sion, which may be the ultimate source of Sr and F. This Table 4 U–Th isotopic data for single calcite bands of Castelnuovo dell’Abate

TRS1_12, 20 and TRS2/2_35/90 refer to calcite bands drilled along transects Sample ID U (mg kg−1₎ 232_Th (μg kg−1₎ ( 230_Th/232_Th) ₍230_Th/238_U) _±2σ ₍234_U/238_U) _±2σ ₍230_Th/238_{U) age} (ka) ±2σ Initial (234_U/238_U) TRS1_12 2.84 782.0 26.64 2.4221 0.0175 2.0975 0.0060 385 75 4.5273 TRS1_20 2.34 1041 16.72 2.4517 0.0119 2.1346 0.0054 368 112 4.6568 TRS2/1_25 2.10 136.9 114.55 2.4587 0.0102 2.1238 0.0032 389 19 4.4317 TRS2/2_35 1.00 18.07 405.0 2.4110 0.0168 2.1046 0.0040 370 14 4.1615 TRS2/2_90 6.73 36.95 1557.8 2.8170 0.0131 2.1411 0.0043 Unable to calculate Unable to calculate

(16)

interaction, however, did not yield very saline fluids as tes-tified by the salinity observed in fluid inclusions (maximum salinity = 1.4 wt% NaCl eq.) (see “Calcite vein deposition” section), likely because Triassic evaporites in southern Tuscany are mainly composed by anhydrite (or gypsum), which is not extremely soluble, whereas halite occurs only in minor amounts (De Paola et al. 2008).

Arsenic is frequently reported in thermal springs, although its occurrence in travertines is rarely investigated (Pentecost 2005). Arsenic, however, is an important com-ponent in travertine deposits of Western Turkey (Dogan and Dogan 2007), Greece (Winkel et al. 2013) and Italy (Dessau 1968; Di Benedetto et al. 2006; Costagliola et al.

2013). The association between travertine and As in south-ern Tuscany and northsouth-ern Latium is well documented, and up to hundreds mg kg−1_{of As is found in fossil travertine}

and calcareous tufa deposits (Costagliola et al. 2010, 2013; Di Benedetto et al. 2011). At Castelnuovo dell’Abate, As is mainly associated with primary pyrite (and pseudomorph Fe-(hydr)oxides after pyrite), which possibly scavenged the metalloid from the hydrothermal fluids.

Calcite vein deposition

Temperature and salinity of the fluid that formed the Castelnuovo dell’Abate banded travertine can be obtained from the fluid inclusion data. Fluid inclusions petrography indicates that several inclusions formed during fluid immis-cibility, consequently, Th of such inclusions corresponds to the trapping temperature (e.g. Shepherd et al. 1985). More-over, considering the shallow environment of travertine for-mation characterized by fluid pressure not exceeding few tens of bars (see later in this paragraph), pressure correc-tion to be added to Th to obtain the trapping temperature is insignificant, and Th values of all L1 inclusions can be considered representative of the trapping temperatures.

As different FIAs are often characterized by distinct Th ranges, temperature fluctuations probably occurred during calcite deposition. Such fluctuation may have produced thermal re-equilibration of some fluid inclusions, specifi-cally of those occurring in the few FIAs that showed not consistent data accordingly to the criteria of Goldstein and Reynolds (1994) and possibly of single large fluid inclu-sions. Moreover, because of the scarce and random dis-tribution of fluid inclusions in calcite, microthermometric data do not allow specific characterization of temperature and salinity for different calcite generations but provide general information during calcite precipitation. Therefore, as fluid inclusions homogenized between 90 and 140 °C, the temperature of calcite forming solution was at least within this range.

According to field observations, the Castelnuovo dell’Abate banded travertine occurs as multiple generations

of banded calcite veins, which propagated from a feed-ing channel localized along a tract of the Castelnuovo dell’Abate fault. The fault acted as a preferential conduit for the ascending Ca-rich fluids, which came to the sur-face precipitating travertine. Accordingly, thermal springs occurring in southern Tuscany and in other geothermal fields elsewhere mainly relate to extensional structures (Kerrich 1986; Hancock et al. 1999; Brogi and Capez-zuoli 2009), enhancing the genetic link between travertine deposition and recent tectonics. The local occurrence of wall rock fragments in the banded calcite veins suggests that travertine deposition was accompanied by the injec-tion of overpressured fluids, probably associated with a hydrothermal eruption (cf., Hedenquist and Henley 1985) and hydrofracturing (Gudmundsson et al. 2002; Uysal et al.

2009; Buttinelli et al. 2011). Recent episodes of fossil and present-day hydrofracturing are documented in the nearby Larderello geothermal area, in the form of phreatic craters and explosions (Marinelli 1969), driven by high-pressure fluids (Ruggieri and Gianelli 1999; Gianelli 2008).

The above-mentioned features as well as the banded vein texture suggest a rhythmical precipitation of calcite caused by cyclic events that can be schematized in five steps: (a) light depressurization of the hydrothermal system by hydraulic fracturing, possibly triggered by tectonic activ-ity; (b) immiscibility of a liquid and vapour phase for adi-abatic decompression (Fournier 1985); (c) ascent of the fluid in the near-vertical channel; (d) CO₂ degassing and associ-ated calcite (and then travertine) deposition by the reaction Ca(HCO3)2 → CaCO3 + CO2 (↑) + H2O (Brasier 2011); and (e) sealing of the conduits that reduced the rock permeability and prepared the system for a new cycle. Under this scenario, the multiple calcite generation as well as their banded fabric is indicative of a travertine multistage deposition.

The relative wide range of variability of apparent salini-ties in fluid inclusions (0.0–1.4 wt% NaCl eq.) might pro-vide further information on the chemical composition of the parent fluid. However, the possible presence of dis-solved CO₂ in fluid inclusions should be considered, as this may depress the Tm_ice and thus result in an overesti-mation of the real salinity (Hedenquist and Henley 1985). In L1 inclusions, the range of CO2 varies between 0 mol% (inclusions displaying Tm_ice of 0.0 °C), suggesting that they represent almost pure water, and 0.14 mol%, as indi-cated by the crushing tests. The variable amount of CO₂ can be easily explained considering the cyclic evolution of the systems precipitating calcite: the self-sealing pro-cess produces the closing of the system towards the surface and the increase in pressure above hydrostatic conditions, causing the system to reach its maximum CO2 content. The maximum pressure reached in the system can be evaluated from the Duan and Sun (2003) CO₂ solubility data, assum-ing that the maximum CO₂ content was reached when the

(17)

fluid was at 140 °C with a salinity of about 1.4 wt% NaCl eq. (the maximum value of salinity observed). Under these conditions, the fluid pressure was ~13 bars. Hydraulic frac-turing led then to depressurization and, consequently, to fluid immiscibility with CO₂ degassing. The occurrence of fluid inclusions that homogenize below 100 °C is further compatible with a hydrothermal solution encompassing cyclic cooling (due to boiling) followed by a new heating and eventually a new heterogeneous trapping.

The highest CO2 content in the trapped fluid (0.14 mol%) gives a Tm_ice depression of 0.15 °C, cor-responding to a maximum overestimation of the salin-ity computed from Tm_ice of 0.27 wt% NaCl eq. Thus, the salinity range showed by L1 inclusions can be only mar-ginally ascribed to changes in the CO₂ content of the fluid. Different salinities in fluid inclusions may be indicative of one or more of these three processes occurring during trap-ping: (1) mixing of two fluids, characterized by different salinities but comparable temperatures; (2) condensation of the low-salinity vapour resulting from fluid immiscibil-ity processes; and (3) immiscibilimmiscibil-ity taking place under dis-equilibrium conditions for CO2.

The depth of the Castelnuovo dell’Abate system can be estimated assuming that the maximum pressure of the fluid (13 bars) corresponds to the pressure necessary to produce hydrofracturing. In extensional tectonic regimes, such as in southern Tuscany, the following relation proposed by Hub-bert and Willis (1957) may apply:

Since the P_lithostatic is approximately 2.5 times the P hy-drostatic, Phydrostatic was estimated to be about 9 bars corre-sponding to a depth of travertine formation of about 75 m. Such a depth is in agreement with the difference in alti-tude between the roof of bedded travertine and the base of banded veins, also considering the erosion that the surface deposit has undergone in 350–400 ka.

Isotopic composition and age

The 13_C/12_{C and}18_O/16_{O isotopic ratios of Castelnuovo} dell’Abate calcite banded veins are in the low range of those observed for hot-spring travertines of central Italy (Fig. 13), and similar to the isotopic ratios commonly observed for thermal-derived deposits (Pentecost 2005; Turi 1986).

Stable C and O isotopes have proved to be a suitable tool to understand the genesis of travertine (Friedman 1970; Manfra et al. 1976; Turi 1986), and the provenance of its source fluid (Minissale et al. 2002). Theoretically, informa-tion on the fracinforma-tionainforma-tion factors in the CO₂–CaCO₃–H₂O system allows to the calculation of the δ18_{O and δ}13_{C of the}

Phydrofracturing∼= (Plithostatic

+_2P_hydrostatic) 3

parent solution from the measurable ratios in fossil traver-tine. Isotopic equilibrium, however, is rarely attained dur-ing travertine deposition (Gonfiantini et al. 1968; Friedman

1970), since kinetic effects (fluid degassing and evapora-tion, microbial activity) and diagenesis are highly effec-tive in causing appreciable shifts from the theoretically expected isotope composition (Friedman 1970; Kele et al.

2008; Manfra et al. 1974; Minissale et al. 2002; Pentecost

2005).

For C, fractionation occurs mainly during CO2 degas-sing, resulting in loss of light 12_{C and consequent increase} of δ13_{C in the deposited rock (Guo et al.}₁₉₉₆_{). In contrast,} CO₂ degassing does not significantly affect δ18_{O of} dis-solved C compounds, which are generally buffered trough rapid re-equilibration by the O isotope composition of the water (Zheng 1990). Oxygen composition, however, is in turn influenced by the preferential evaporation of H₂16_O (Turi 1986). Consequently, downstream of the spring, trav-ertines progressively become enriched in 18_{O and}13_C, dis-playing a linear positive δ18_O–δ13_{C correlation (Gonfiantini} et al. 1968; Kele et al. 2008). This trend can be expected to be quite strong for surface (open air) travertine precipita-tion (e.g. cascades or lacustrine facies), but relatively weak precipitation along veins, such as Castelnuovo dell’Abate banded travertine. Here, the absence of a clear positive δ18_O– δ13C correlation (Fig. 14) (Vermoere et al. 1999) and the fairly constant C–O isotope composition of calcite (Table 3) suggest that: (a) the isotopic composition of waters remained constant during travertine deposition, and (possibly) traver-tine precipitated under isotopic equilibrium, or (b) a number of factors (T, isotopic composition of the fluid, Rayleigh dis-tillation processes, etc.) varied in the same way, resulting in negligible changes in the δ18_O–δ13_{C systematics.}

Assuming equilibrium conditions between fluid and calcite, the δ18_{O composition of the parent fluid was} deter-mined by the equation of Friedman and O’Neil (1977), supposing that most of calcite deposition had occurred in a temperature range of 100–140 °C (Fig. 10). The result-ing δ18_{O of the hydrothermal water ranges from +0.9} to +7.4 ‰ (Fig. 15) and is strongly enriched in 18_{O with} respect to the present-day regional rainwater (δ18_{O = −7.0} to −6.0 ‰; Fig. 15; Longinelli and Selmo 2003), which is believed to feed the shallow hydrothermal systems of south-ern Tuscany (Minissale et al. 1997; Minissale 2004). Even though some isotopic shift due to disequilibrium is con-ceivable, a strictly meteoric composition for the travertine fossil fluid is unrealistic. The O composition of a hypotheti-cal hypotheti-calcite deposited in a temperature range of 100–140 °C from a typical meteoric fluid (δ18_{O = −7.0 ‰) would} have resulted in δ18_{O ranging from 6 to 10 ‰. Therefore,} to account for the δ18_{O values measured in Castelnuovo} dell’Abate calcites (δ18_{O = +18.1 to +21 ‰), an isotopic} shift of more than 10 ‰ should be assumed, which is much

(18)

higher than that observed for deposits formed on the sur-face under disequilibrium conditions (Kele et al. 2008,

2011). The simplified assumption of travertine deposition under equilibrium conditions is believed then to be valid, at least to indicate that the Castelnuovo dell’Abate fos-sil fluid was enriched in 18_{O (likely caused by water–rock} interaction processes, see “Present and fossil hydrothermal systems of southern Tuscany” section) with respect to the present-day regional rainwater.

Carbon isotopic composition of CO₂ (δ13_C

CO2) was

calculated using the equation of Bottinga (1969), which directly relates the δ13_{C of fossil deposits (δ}13_C

CaCO3)

to δ13_C

CO2. Additionally, δ

13_C

CO2 was derived using

the empirical equation of Panichi and Tongiorgi (1976) (δ13_C

CO2 = 1.2 δ

13_C

CaCO3—10.5), obtained specifically for

Italian travertine.

For the assumed temperature range of deposition (100–140 °C), the calculated δ13_C

CO2 ranged from −3.2 to +1.2 ‰ (Bottinga 1969) and from −9.5 to −7.4 ‰ (Pan-ichi and Tongiorgi 1976), respectively. The composition derived from Panichi and Tongiorgi (1976) is then mark-edly more negative than that determined by the Bottinga’s equation. Such large difference is probably related to the different conditions that apply to the two equations: the Bottinga (1969) equation requires equilibrium conditions, whereas the empirical relationship of Panichi and Tongiorgi (1976) was obtained for travertine deposited in surface environments characterized by large and rapid degassing processes which strongly affect δ13_C

CO2. Bottinga’s

equa-tion appears then to be appropriate to calculate the δ13_C CO2

of Castelnuovo dell’Abate fluids since this banded traver-tine formed under hypogenetic conditions, and significant degassing occurred only during hydraulic fracturing events and consequent depressurization. Moreover, the δ13_C

CO2

of the nearby travertine deposits of Bagni San Filippo and Bagno Vignoni (−6.1 to −2.0 ‰; Fig. 1), calculated using the Panichi and Tongiorgi (1976) equation from δ13_C

CaCO3

(+3.4 to +6.8 ‰; Minissale 2004), is systematically higher than those calculated for the Castelnuovo dell’Abate sys-tem with the same equation, whereas the δ13_C

CO2 values of

the three sites partially overlap when calculated using the Bottinga (1969) equation. Thus, even if the assumption of travertine deposition under equilibrium conditions may be not totally respected, the δ13_C

CO2 range obtained from

Bottinga (1969) is to some extent similar to the δ13_C CO2

range (−4.6 to −2.3 ‰; Minissale 2004; Frondini et al.

2009) of the present-day CO₂-rich gas emissions occurring in the area surrounding Castelnuovo dell’Abate. The CO2 in these emissions and more in general in the emissions of the Monte Amiata area is believed to result from a mixture of different sources such as mantle degassing and thermo-metamorphic (decarbonation) reactions (Minissale 2004; Frondini et al. 2009; Tassi et al. 2009). A similar composite origin can be proposed also for the CO₂ of the Castelnuovo dell’Abate, although the more positive δ13_C

CO2 values

dis-played by this system suggest a larger contribution from decarbonation reaction and/or 13_{C enrichment due to} degas-sing as a consequence of hydraulic fracturing events.

The widespread and sustained hydrothermal activity in the Monte Amiata area is commonly related to magma-tism, whose volcanic products have radiometric ages of 300–200 ka, based on K/Ar (Bigazzi et al. 1981; Cadoux and Pinti 2009) and 40_Ar/39_{Ar (Laurenzi and Villa}₁₉₉₁_; Barberi et al. 1994) dating of sanidine and plagioclase, and fission tracks on volcanic glass (Bigazzi et al. 1981). To date, the only available age data for the hydrothermal circu-lation related to the Monte Amiata thermal anomaly is the 230_Th/234_{Th age of some lacustrine deposits (diatomites)} (Fornaca Rinaldi 1968). The 370–390 ka age data obtained for calcite in this study provide clear evidence that the hydrothermal systems were already established before the Monte Amiata volcanic manifestations, suggesting that the thermal anomaly at Monte Amiata predated volcanism. This observation agrees with the model proposed by Cadoux and Pinti (2009), which suggests a long-pre-eruptive evolu-tion of the magma chamber of the Monte Amiata volcano. According to this model, the thermal anomaly developed in response to the emplacement of a large magma body in the upper crust. Such intrusion probably corresponds to the large uplifted region in the Monte Amiata area, showing an elliptical shape with major axes of 25–30 and 45–50 km (Gianelli et al. 1988). Ages up to 218 ka have been reported for the travertine plate of Montemerano (Fig. 1; Taddeucci Fig. 15 Homogenization temperatures versus δ18_{O for the banded}

calcite veins of Castelnuovo dell’Abate. The δ18_{O composition of the}

fluid in equilibrium with calcite was calculated using the fractiona-tion equafractiona-tion of Friedman and O’Neil (1977), in the 100–140 °C Th range. The composition of Bagni San Filippo modern and fossil ther-mal springs and of rainwater of southern Tuscany is reported. *Data from Gasparrini et al. (2013); **from Fancelli and Nuti (1975), DST (2010)

(19)

and Voltaggio 1987), and the stratigraphic age of the nearby Semproniano travertine (Fig. 1) has been assigned to the Lower Pliocene (Bosi et al. 1996). These deposits further document the presence of a thermal anomaly that predates the Monte Amiata volcanism.

As suggested by this study, at least part of the CO2 of the shallow hydrothermal system at Castelnuovo dell’Abate had a deep source, whereas the low salinities of fluid inclu-sions exclude a direct input of magmatic-derived water in the system. It is conceivable that the effect of the magma emplacement at Monte Amiata was limited to the influx of deep CO₂ and to the activation of a hydrothermal circula-tion, which was in turn fed by meteoric waters.

Present and fossil hydrothermal systems of southern Tuscany

Abundance of hot springs, CO2-dominated gas emissions, travertine-depositing manifestations and the widespread occurrence of CO₂ in present-day geothermal fields pro-vide clear epro-vidence for active CO₂-rich hydrothermal sys-tems occurring in southern Tuscany (Minissale et al. 1997). Bagni San Filippo is one of the best characterized examples of long-lived (since the Pleistocene) CO₂-rich hydrother-mal systems in the region, as suggested by the presence of an exhumed Hg mineralization with Pleistocene age. The water currently discharged by this system, depositing trav-ertine on the surface, is characterized by TDS of 4128 mg/l and a temperature of 52 °C (Minissale 2004). Equilibrium temperatures of the aquifer feeding the surface system are comprised in the 60–95 °C range, whereas pCO₂ value esti-mated at 60 °C is about 10 bar (Donnini et al. 2007, Baietto et al. 2008).

In the following discussion, the travertine deposit asso-ciated with the thermal spring of Bagni San Filippo (and Bagno Vignoni) will be considered as a modern analogue of the Castelnuovo dell’Abate system since these locali-ties are located within a radius of few kilometres of dis-tance and share some common geological and geochemi-cal features. Hydrothermal circulation at these sites was favoured by the presence of Neogene-Quaternary brittle structures (Brogi et al. 2010a, b, 2012) that allowed the upwelling of large volumes of meteoric-derived aqueous solutions, heated by the regional high geothermal gradient and enriched in Sr and F through their circulation in Mes-ozoic carbonate and Triassic evaporite successions (e.g. host rocks of the shallower reservoir; Duchi et al. 1992; Cortecci and Lupi 1994; Gasparrini et al. 2013; Brogi et al. 2015).

Following the data of Gasparrini et al. (2013) on hydrothermal calcite veins sampled in Mesozoic carbon-ate rocks of the Monte Amiata surroundings, the Bagni San Filippo fossil hydrothermal system was characterized

by temperatures of about 165 °C, relatively low salinities (generally comprised between 1.2 and 1.7 wt% NaCl eq.), δ13C = –2.6 to +0.6 ‰ and δ18O = + 21.6 to +23.1 ‰ (Gasparrini et al. 2013) (Fig. 10). These isotopic values are similar to those observed in the Castelnuovo dell’Abate fossil system (Table 3), further highlighting the similarities between these two systems.

The present-day hydrothermal system of Bagni San Filippo shows a lighter isotopic composition of water with δ18_{O between –8.2 ‰ and –4.8 ‰ (Gonfiantini et al.}

1968; Fancelli and Nuti 1975; DST 2010; Fig. 15), thus markedly different from the (calculated) much heavier O isotope composition of the water that fed the same system in the past (δ18_{O = + 8.4 to +11.6 ‰; Gasparrini et al.}

2013) and from the water that deposited the Castelnuovo dell’Abate travertine (δ18_{O = + 0.9 to +7.4 ‰; this} study). Accordingly, hydrothermal calcite veins devel-oped around the geothermal anomalies of southern Tus-cany generally indicate an O composition in fossil fluids heavier than meteoric water (Liotta et al. 2009). Concern-ing the Castelnuovo dell’Abate and Bagni San Filippo fossil hydrothermal systems, inputs of heavy O isotopes from a magmatic source water could be reasonably excluded based on the low salinities of the fluid inclusions in calcite, which suggest a meteoric origin of the origi-nal waters feeding the Castelnuovo dell’Abate system. In contrast, it could be argued that in fossil hydrothermal systems, the meteoric waters interacted, more extensively than today, with the rocks belonging to the Tuscan Nappe rock pile (Cortecci and Lupi 1994; Gasparrini et al. 2013) and relatively enriched in 18_{O. The O isotopic signatures} of calcite of the Tuscan succession are markedly positive (δ18_{O ~ +25 ‰), with values as high as +28.9 ‰ for the} Jurassic limestone (Calcare Massiccio; Fig. 2) (Cortecci and Lupi 1994). This feature marks a clear difference between the fossil and the modern surficial hydrother-mal system at least in the Castelnuovo dell’Abate-Bagni San Filippo area. 18_{O enrichment by water–rock} interac-tion, however, is evident from the isotopic composition (δ18_{O = –3 to +1.8 ‰, DST}₂₀₁₀_{) of the fluid produced} from the shallow reservoir hosted in carbonate rocks of the Monte Amiata geothermal field.

According to fluid inclusions data on hydrothermal calcite veins (Gasparrini et al. 2013), the fluid reser-voir of Bagni San Filippo experienced a cooling of about 70–105 °C from vein formation (Pleistocene) to mod-ern time. A similar trend has been observed in the deep Larderello geothermal reservoir, which underwent to a temperature decrease of about 120 °C in the last 3 Ma (Del Moro et al. 1982). Assuming that the Castelnuovo dell’Abate system had the same temperature of the Bagni San Filippo travertine-depositing spring (~50 °C) when it was active, and taking into account the calcite ages, it